No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Use of routine clinical laboratory data to define reference intervals
Abstract
Reference intervals are used to distinguish between healthy and diseased state. Ideally, they are defined using specimens only from 'healthy' individuals, but this is often difficult or impossible. In order to use routine clinical laboratory data, outliers must be removed before the underlying distribution and changes related to age and sex can be modelled. This paper illustrates the process for plasma alkaline phosphatase (ALP). ALP levels are high in infancy and childhood, peak in adolescence, are stable from the early 20s and rise after the fourth decade. Three types of normalizing transformations (Logarithmic, Box-Cox and Cole's LMS) are compared.
Single ALP results from 75,328 individuals aged 0-80 years were binned by sex and age. The normalizing transformations were applied to each bin, outliers were removed and the normalizing transformations were reapplied to the remaining data. The normality of the transformed data was assessed by normal score plots and the Kolmogorov-Smirnov test. Fractional polynomials were used to model the underlying parameters of the transformations and the derived parametric reference intervals (mean +/- 1.96 standard deviations), separately for each sex as a whole and partitioned into two or three age ranges, with overlapping to give smooth transitions.
All transformations yielded acceptably normal data, but the LMS method gave the closest approximation to normal. Outlier rates were similar for each method. The derived reference ranges were similar for all the three methods. Splitting the data-set into several segments resulted in a better fit with the peak seen in adolescence.
Routine clinical laboratory specimens can be used to derive reference intervals.
... Reference intervals (RIs) are typically derived from historical literature values, laboratory instrument manufacturers or appropriate statistical analysis of routine clinical laboratory patient data, all of which can introduce sampling bias. 1,2 Blood results are affected by individual and laboratory factors, such as differences in age, gender, ethnicity, laboratory methods and laboratory instruments. [3][4][5] While the gold standard is to minimize laboratory and population differences by computing institutional RIs, there are inherent problems with obtaining a sufficient sample size for such analyses, so validation of externally obtained RIs is recommended. ...
... Discrepancies between the study population and the institutional RIs were further evaluated by using a well-recognized method for defining RIs, which is the basis for current guidelines. 2,6,55,56 The central 95% of results for each analyte, subgrouped by gender, were calculated following removal of outliers. This method aims to maximise sensitivity and specificity by obtaining information from the extremes of the sample, while avoiding outliers adversely influencing the analysis. ...
... aLP aLP levels are stable in young adults, following cessation of growth and bone turnover and increase in the fourth decade. 2,40 aLT BMI has the strongest independent association with aLT in healthy populations, but increased aLT levels are also associated with male gender age, peaking at 55 y of age. 4,41,42 Non-alcoholic fatty liver disease (NaFLD) is an important cause of persistently elevated aLT levels in asymptomatic adults and is estimated to have a prevalence of 20-30% in countries such as the UK. ...
This was a retrospective study to determine the validity of institutional reference intervals for interpreting biochemistry and hematology results in healthy adults in the context of clinical trials of preventive vaccines. An example population of 974 healthy adults participating in clinical trials at the Jenner Institute, Oxford, UK, between 1999 and 2009 was studied. Methods for calculating the central 95% ranges and determining the coefficients of within person variation were demonstrated. Recommendations have been made as to how these data can be usefully applied to the interpretation of blood results in healthy adult subjects for the purposes of clinical trial inclusion decisions and post-vaccination safety monitoring.
... The modified Box-Cox transformation [36,37] is also commonly used and is considered a more flexible transformation method [38]. However, there is often a risk of incorporating prevalence data this way, leading to the inadvertent widening of RIs [39,40]. Other non-Gaussian transformation methods such as Manly transformation and Erlang1 distribution [20] are also available, but care must be taken when performing data transformation to avoid over-transformation, which can cause the RI to lose specificity [23]. ...
Reference intervals (RIs) are the cornerstone for evaluation of test results in clinical practice and are invaluable in judging patient health and making clinical decisions. Establishing RIs based on clinical laboratory data is a branch of real-world data mining research. Compared to the traditional direct method, this indirect approach is highly practical, widely applicable, and low-cost. Improving the accuracy of RIs requires not only the collection of sufficient data and the use of correct statistical methods, but also proper stratification of heterogeneous subpopulations. This includes the establishment of age-specific RIs and taking into account other characteristics of reference individuals. Although there are many studies on establishing RIs by indirect methods, it is still very difficult for laboratories to select appropriate statistical methods due to the lack of formal guidelines. This review describes the application of real-world data and an approach for establishing indirect reference intervals (iRIs). We summarize the processes for establishing iRIs using real-world data and analyze the principle and applicable scope of the indirect method model in detail. Moreover, we compare different methods for constructing growth curves to establish age-specific RIs, in hopes of providing laboratories with a reference for establishing specific iRIs and giving new insight into clinical laboratory RI research. (201 words).
... Los datos de laboratorio existentes pueden utilizarse para establecer Intervalos de referencia siempre y cuando se indague el estado de salud de la población estudiada, para esto se pueden utilizar métodos estadísticos que ayudan a determinar valores atípicos. 24,25 Una ventaja de este enfoque es la capacidad de acumular un gran número de individuos para cada clasificación por edad y sexo. ...
Introducción los resultados de laboratorio clínico deben interpretarse a la luz de intervalos biológicos obtenidos de individuos de referencia, en estos juega un papel muy importante la variación intra e interindividual de las magnitudes biológicas y de factores como la nutrición y el origen geográfico, entre otros. Dichos valores generalmente son calculados con herramientas estadísticas a las que no se les comprueba los supuestos estadísticos, o no se tiene en cuenta el tamaño muestral requerido afectando su validez. Métodos: este estudio utilizó los métodos percentil empírico, Bootstrap, Harrell & Davis, y el estimador robusto de Horn de acuerdo a la clasificación por edad y sexo recomendada por Soldin et al. en 2003 en población pediátrica, para estimar intervalos de referencia biológicos de 20 parámetros del hemograma de 842 personas entre 2 y 18 años. Los intervalos propuestos de los siguientes mensurandos: leucocitos, glóbulos rojos, hemoglobina y plaquetas, fueron comparados frente a los propuestos por dichos autores, para determinar si habían cambios sustanciales en las poblaciones estudiadas y establecer cuál de los métodos evaluados determinaría un Intervalo con el que el clínico pudiera apoyar el estado hematológico de la población estudiada. Resultados y conclusiones : los métodos Bootstrap y estimador robusto de Horn producen intervalos más amplios que el estimador de Harrell & Davis y percentil empírico, en la mayoría de los casos. El límite inferior calculado con el estimador Robusto de Horn se alejó mucho más del valor de la mediana, y el método Bootstrap produjo el límite superior más alto. Al comparar cada intervalo con los propuestos por Soldin et al., se observaron diferencias marcadas aun cuando la variabilidad intraindividual fue baja como en el caso de la hemoglobina. Estos resultados confirman la necesidad de que cada laboratorio estime sus propios intervalos biológicos utilizando protocolos estandarizados.
... Finally, WBC may be elevated in women with non-infectious inflammatory conditions. Previous studies have demonstrated that, with a suitably large population, RIs can be accurately estimated in this way from unselected populations [36]. Other methods may be used to limit the selection criteria further, including using proxy biomarkers (e.g., haemoglobin or platelets) to identify women with other haematological conditions. ...
Background
White blood cells (WBC) are commonly measured to investigate suspected infection and inflammation in pregnant women, but the pregnancy-specific reference interval is variably reported, increasing diagnostic uncertainty in this high-risk population. It is essential that clinicians can interpret WBC results in the context of normal pregnant physiology, given the huge global burden of infection on maternal mortality.
Methods
We performed a longitudinal, repeated measures population study of 24,318 pregnant women in Oxford, UK, to map the trajectory of WBC between 8-40 weeks of gestation. We defined 95% reference intervals (RI) for total WBC, neutrophils, lymphocytes, eosinophils, basophils, and monocytes for the antenatal and postnatal periods.
Findings
WBC were measured 80,637 times over five years. The upper reference limit for total WBC was elevated by 36% in pregnancy (RI 5.7-15.0×109/L), driven by a 55% increase in neutrophils (3.7-11.6×109/L) and 38% increase in monocytes (0.3-1.1×109/L), which remained stable between 8-40 weeks. Lymphocytes were reduced by 36% (1.0-2.9×109/L), while eosinophils and basophils were unchanged. Total WBC was elevated significantly further from the first day after birth (similar regardless of the mode of delivery), which resolved to pre-delivery levels by an average of seven days, and to pre-pregnancy levels by day 21.
Interpretation
There are marked changes in WBC in pregnancy, with substantial differences between cell subtypes. WBC are measured frequently in pregnant women in obstetric and non-obstetric settings, and results should be interpreted using a pregnancy-specific RI until delivery, and between days 7-21 after childbirth.
... [49]. Simple logarithmic transformation is usually all that is required, and Box-Cox methods usually provide better normalisation of the data but the improvements are usually small [50], and come with the risk of including diseased individuals in the reference population. For example, the typically log normal distribution of alkaline phosphatase (ALP) might be inadvertently extended to include patients with vitamin D deficiency [51]. ...
The indirect approach to defining reference intervals operates ‘ a posteriori ’, on stored laboratory data. It relies on being able to separate healthy and diseased populations using one or both of clinical techniques or statistical techniques. These techniques are also fundamental in a priori , direct reference interval approaches. The clinical techniques rely on using clinical data that is stored either in the electronic health record or within the laboratory database, to exclude patients with possible disease. It depends on the investigators understanding of the data and the pathological impacts on tests. The statistical technique relies on identifying a dominant, apparently healthy, typically Gaussian distribution, which is unaffected by the overlapping populations with higher (or lower) results. It depends on having large databases to give confidence in the extrapolation of the narrow portion of overall distribution representing unaffected individuals. The statistical issues involved can be complex, and can result in unintended bias, particularly when the impacts of disease and the physiological variations in the data are under appreciated.
... 21 Data on NT-pro BNP were approximated to a normal distribution in each trimester using a logarithmic transformation. Participants were binned into equal groups of maternal age and body mass index (BMI) for each biomarker, as performed in other studies of RIs, 22 and outliers were identified and excluded within these bins using Horn's method. 23 BNP could not be transformed so outliers were identified by visually examining the distribution. ...
Background
Cardiac disease is the leading cause of maternal mortality in the UK, so accurate cardiovascular diagnoses in pregnancy are essential. BNP (B-type natriuretic peptide) and NT-pro BNP (N-terminal-pro BNP) are useful clinical tools for investigating suspected peripartum cardiomyopathy but, as the pregnancy-specific reference intervals are undefined, it is uncertain how they should be interpreted in pregnant women.
Methods
Longitudinal study of 260 healthy pregnant women, with sampling in each trimester to define 95% reference intervals.
Results
The upper reference limit for NT-pro BNP was 200 pg/mL in the first and second trimesters, and 150 pg/mL in the third. Levels were significantly reduced in overweight women in the third trimester (p=0.0001), which supports the partitioning of reference intervals by BMI. The upper limit for BNP was 50 pg/mL, with no detectable trimester-related differences. Whilst other biomarkers (haemoglobin and platelets) fell throughout pregnancy, both natriuretic peptides were initially elevated before falling by the third trimester, suggesting that the observed changes in natriuretic peptides are driven by dynamic interplay between cardiac strain and progressive haemodilution. NT-pro BNP in the first trimester was inversely associated with neonatal birthweight at term (p=0.011).
Conclusions
Cardiac biomarkers have an important role for investigating suspected disease in high-risk pregnant women, but a robust assessment of the levels expected in healthy pregnant women is an essential prerequisite to their application in clinical practice. This study has defined trimester- and BMI-specific reference intervals for NT-pro BNP and BNP, which may improve how women with suspected cardiovascular disease are investigated in pregnancy.
... Este es un factor crucial de los métodos indirectos. En los diferentes estudios descritos en la literatura [24,[29][30][31][32][33][34][35][36][37][38] los métodos estadísticos empleados se pueden agrupar en dos estrategias: -Grupo A: Basándonos en el conjunto de datos, se aplican técnicas estadísticas para eliminar datos extremos o atípicos (outliers), antes de emplear otros métodos estadísticos para calcular los intervalos de referencia. -Grupo B: Se aplica directamente un método estadístico a todo el conjunto de datos sin eliminar ninguno previamente para calcular los intervalos de referencia. ...
Resumen
Los intervalos de referencia son habitualmente empleados como herramienta de apoyo a las decisiones clínicas. En esta revisión se resumen los aspectos relacionados con el big data y los intervalos de referencia, las prácticas actuales, incluyendo los métodos estadísticos, los requisitos de calidad de los datos, incluyendo la armonización y la normalización, y las perspectivas de futuro para la determinación indirecta de intervalos de referencia mediante datos de laboratorio de rutina.
... This is a critical point in indirect studies. In the different projects that have been described in the literature [24,[29][30][31][32][33][34][35][36][37][38], the statistical methods used can be grouped into two main data management strategies: -Group A: Based on the data set, statistical techniques for the elimination of extreme or atypical data (outliers) are applied, before using other statistical procedures to calculate the reference intervals. -Group B: Directly applies statistical methods over the entire data collection, without eliminating any of them by means of atypical value detection techniques for the calculation of the reference intervals. ...
Reference intervals are commonly used as a decision-making tool. In this review, we provide an overview on “big data” and reference intervals, describing the rationale, current practices including statistical methods, essential prerequisites concerning data quality, including harmonization and standardization, and future perspectives of the indirect determination of reference intervals using routine laboratory data.
... Generally speaking, RIs are reported as population-based values comprising 95% of the healthy population in direct studies for RI estimation [3,4]. RIs are typically derived from historical research values, instrument manufacturers, or suitable statistical analysis of patient data, all of which may result in sampling bias [5,6]. They are usually affected by individual and laboratory factors, such as differences in age, gender, ethnicity, region, environment, genetic factors, laboratory methods, and instruments [7][8][9][10]. ...
PurposeLaboratory reference intervals (RIs) play a key role in clinical pharmacology trials, both in the screening process and in evaluating drug safety. However, RIs tend to be confined to the general population, and data about RIs for the trial population are limited. The purpose of this study was to determine appropriate RIs for use when screening a defined special subgroup of a healthy Chinese population in clinical pharmacology trials.MethodsA total of 773 healthy Chinese volunteers (552 men and 221 women) who sought to participate in clinical pharmacology trials were included in this study. Sixteen different biochemical analytes were measured by a Beckman Coulter Unicel DxC 800 automatic analyzer. RIs were partitioned by gender using Harris and Boyd’s method and calculated using a non-parametric method.ResultsThe RIs of 16 biochemical analytes for healthy Chinese volunteers during the screening process in clinical pharmacology trials are reported in this study. Noticeable differences between the RIs in this study and RIs provided by our laboratory or existing literature were also observed. Compared to our institutional RIs, the newly established RIs were more applicable to the current trial population.Conclusions
The RIs in this study can serve as a powerful clinical tool during the screening process in clinical pharmacology trials. However, these RIs should be re-verified if any condition changes. The results also emphasize the importance re-establishing RIs which are more applicable to local trial populations.
... One of the strengths of this study is that in comparison to previously reported RIs with limited partitioning into large-scale age groups, our study demonstrates a more fine-grained representation of changes in ALP activity with age. [14][15][16][17][18] The Siemens Advia 1800 manufacturer package insert RI, that is, 46 to 116 IU/L, lacks appropriate portioning according to the age and sex and differs significantly from the RIs established by this study. Using RIs for clinical decision making based on limited partitioning into large-scale age groups can only approximate the true physiological dynamics of ALP activity. ...
Objective:
To establish reference intervals (RIs) for alkaline phosphatase (ALP) levels in Pakistani children using an indirect data mining approach.
Methods:
ALP levels analyzed on a Siemens Advia 1800 analyzer using the International Federation of Clinical Chemistry's photometric method for both inpatients and outpatients aged 1 to 17 years between January 2013 and December 2017, including patients from intensive care units and specialty units, were retrieved. RIs were calculated using a previously validated indirect algorithm developed by the German Society of Clinical Chemistry and Laboratory Medicine's Working Group on Guide Limits.
Results:
From a total of 108,845 results, after the exclusion of patients with multiple specimens, RIs were calculated for 24,628 males and 18,083 females with stratification into fine-grained age groups. These RIs demonstrate the complex age- and sex-related ALP dynamics occurring during physiological development.
Conclusion:
The population-specific RIs serve to allow an accurate understanding of the fluctuations in analyte activity with increasing age and to support clinical decision making.
... These results also accord with previous findings. 9 For the ≥ 90 years age group, IgA seems to increase. However, the dataset had fewer individuals in this age range, and the confidence intervals in Figure 4 overlap, so the trend is not statistically significant. ...
Background: Reference intervals are essential to interpret diagnostic tests, but their determination has become controversial. Methods: In this paper parametric, non-parametric and robust reference intervals with Tukey and block elimination are calculated from a dataset of over 32,000 serum immunoglobulin A (IgA) measurements. Results: The outlier elimination method was significantly more determinative of the reference intervals than the calculation method. The Tukey elimination procedure consistently eliminated significantly more values than the block method of Dixon and Reed across all age ranges. If Tukey elimination was applied, variation between reference intervals produced by the different calculation methods was minimal. Block elimination rarely eliminated values. The non-parametric reference intervals were more sensitive to outliers, which in the IgA context, led to higher and wider reference intervals for the older age groups. There were only minimal differences between robust and parametric reference intervals. Conclusions: This suggests that Tukey elimination should be preferred over the block D/R method for datasets similar to the one used in this study. These are predominantly new observations, as previous literature has focused on the calculation technique and not discussed outlier elimination. This suggests the robust method is not advantageous over the parametric method and therefore due to its complexity is not particularly useful, contrary to CLSI Guidelines.
... Potentially, racial/ethnic-specific RIs will reduce misdiagnosis, over-and under-estimation of disease prevalence rates, the failure or delay in the required reporting of critical laboratory values; 12 however, further work is needed to validate these benefits. Physicians and other healthcare providers use the laboratory test results to track clinical outcomes and make clinical decisions, 37,38 to screen asymptomatic people and to identify those at risk and for early detection of diseases. 39,40 Therefore, accurate RIs for for laboratory tests are important for patients and their caregivers to monitor their health and disease progress. ...
Reference intervals (RIs) for common clinical laboratory tests are usually not developed separately for different subpopulations. The aim of this study was to investigate racial/ethnic differences in RIs of common biochemical and hematological laboratory tests using the National Health and Nutrition Examination Survey (NHANES) 2011-2012 data. This current study included 3,077 participants aged 18-65 years who reported their health status as "Excellent," "Very good," or "Good," with known race/ethnicity as white, black, Hispanic, or Asian. Quantile regression analyses adjusted for sex were conducted to evaluate racial/ethnic differences in the normal ranges of 38 laboratory tests. Significant racial/ethnic differences were found in almost all laboratory tests. Compared to whites, the normal range for Asians significantly shifted to higher values in globulin and total protein and to lower values in creatinine, hematocrit, hemoglobin, mean cell hemoglobin, mean cell hemoglobin concentration, and mean platelet volume. These results indicate that racial/ethnic subpopulations have unique distributions in the labortoary tests and race/ethnicity may need to be incorporated in the development of their RIs. Establishment of racial/ethnic-specific RIs may have significant clinical and public health implication for more accurate disease diagnosis and appropriate treatment to improve quality of patient care, especially for a state with diverse racial/ethnic subpopuations such as Hawai'i.
... It has been adapted to derive biochemistry reference intervals. [15][16][17] It visually presents the biological trend of a parameter with age using smoothed continuous centile lines. The median CV i , shown as 50th centile lines in Figure 2, allow one to appreciate the changes in CV i with age that may be lost if the pediatric subjects are analyzed as a group. ...
... It has been adapted to derive biochemistry reference intervals. [15][16][17] It visually presents the biological trend of a parameter with age using smoothed continuous centile lines. The median CV i , shown as 50th centile lines in Figure 2, allow one to appreciate the changes in CV i with age that may be lost if the pediatric subjects are analyzed as a group. ...
Derivation of between-individual biological variation (CVg) data requires repeat sampling of the same subject, which is undesirable and challenging in children. We describe an indirect sampling (data mining) approach to obtain these data in children.
Twenty-two serum biochemistry results from 6,989 children, who visited their primary care physician in Queensland, Australia, and were tested only twice within a year were included. The CVg and index of individuality of the boys and girls were estimated by year of age, according to the procedures recommended by Fraser and Harris.
The CVg was generally higher during the first year of life and declined to reach a constant level by age 4 to 6 years, except for aspartate aminotransferase, alanine aminotransferase, γ-glutamyltransferase, and phosphate. The CVg for these tended to increase after age 10 years. Most of the serum biochemistries examined in this study had indices of individuality 0.6 or less, except sodium, anion gap, bicarbonate, and chloride, which ranged from 0.6 to 1.4. The indices of individuality were very stable across all ages.
These data are comparable to those reported by the Canadian Laboratory Initiative on Pediatric Reference Intervals study and the Ricos database for adults. This study reports the CVg trends and data for boys and girls by year of age, which have not been described previously.
Copyright© by the American Society for Clinical Pathology.
... It has been adapted to derive biochemistry reference intervals. [15][16][17] It visually presents the biological trend of a parameter with age using smoothed continuous centile lines. The median CV i , shown as 50th centile lines in Figure 2, allow one to appreciate the changes in CV i with age that may be lost if the pediatric subjects are analyzed as a group. ...
Objectives:
Pediatric within-individual biological variation (CVi) is a challenge to derive by direct sampling due to clinical, logistical, and ethical barriers.
Methods:
Laboratory results of 22 basic biochemistry tests performed on 9,356 children who visited primary care physicians more than once over a year were obtained from a large laboratory network in Australia. The CVi were calculated as (CVT (2) - CVa (2))(0.5), where CVT was the coefficient of variation between repeat measurements and CVa was the analytical imprecision. Smoothed 50th centile (median) CVi charts were derived using the LMS ChartMaker Light software (Medical Research Council, Cambridge, England) with L, M, and S parameters fixed at 3.0, 3.0, and 3.0 equivalent degrees of freedom, respectively.
Results:
In general, the median CVi trends for this pediatric cohort remained relatively stable with increasing age. Only aspartate aminotransferase, globulin, phosphate, urea, and creatinine had differences between the highest and lowest median CVi of more than 30%. The differences between the child and adult CVi were relatively small. Nearly all the analytes had child to adult CVi ratios of 1.0 ± 0.5.
Conclusions:
The median CVi derived from patients with only two repeat biochemistry measurements may be considered reasonable estimates of CVi among children seeking treatment at primary care settings. The LMS approach allowed visualization of the continuous trends of CVi with age and extended the pediatric CVi estimation to younger than 4 years.
... We would also require hundreds more for the various stages of childhood 122 which is why CLSI C28-A3 specifically states that indirect techniques 'are used when it is deemed too difficult to collect samples from healthy subjects (e.g. paediatrics)' 13 and investigators have used the indirect approach successfully in both paediatrics 123 and the elderly. 124 It is, therefore, virtually impossible to perform direct reference interval studies with enough individuals to represent all the physiological differences that are known to exist. ...
Reference intervals are ideally defined on apparently healthy individuals and should be distinguished from clinical decision limits that are derived from known diseased patients. Knowledge of physiological changes is a prerequisite for understanding and developing reference intervals. Reference intervals may differ for various subpopulations because of differences in their physiology, most obviously between men and women, but also in childhood, pregnancy and the elderly. Changes in laboratory measurements may be due to various physiological factors starting at birth including weaning, the active toddler, immunological learning, puberty, pregnancy, menopause and ageing. The need to partition reference intervals is required when there are significant physiological changes that need to be recognised. It is important that laboratorians are aware of these changes otherwise reference intervals that attempt to cover a widened inter-individual variability may lose their usefulness. It is virtually impossible for any laboratory to directly develop reference intervals for each of the physiological changes that are currently known, however indirect techniques can be used to develop or validate reference intervals in some difficult situations such as those for children. Physiology describes our life's journey, and it is only when we are familiar with that journey that we can appreciate a pathological departure.
... The reference range is a well-established strategy in medical practice [25,26,27,28,29]. A reference range is simply a confidence interval (generally 95%) for the distribution of test results for a control group, which provides a common-sense framework for interpreting the results of clinical tests. ...
Copy Number Variations (CNVs) are a significant source of human genetic diversity and are believed to be responsible for a wide variety of phenotypic variation. Recent advances in microarray-based genomic hybridization techniques have facilitated CNV analysis as a viable diagnostic technique in the clinic, and several public databases of well-characterized CNVs are being compiled, but a standard for interpreting uncharacterized CNVs has yet to emerge. This thesis examines the clinical interpretation of uncharacterized CNVs as a multiple instance binary classification problem. We analyze the current state of clinical techniques, then present and test several novel statistical approaches to the problem.
... There are numerous challenges with determining these ranges and in using them for clinical decisionmaking . Many factors such as age, sex, and sampling bias can influence these values; it can be difficult to identify healthy individuals; and there is disagreement over which statistical techniques and percentiles to use12131415. Furthermore, it is unclear how useful reference ranges are in clinical decision-making since there is a distinction between a reference limit and the value that will actually change a physician's clinical decision16171819. ...
Evidence-based medicine employs expert opinion and clinical data to inform clinical decision making. The objective of this study is to determine whether it is possible to complement these sources of evidence with information about physician "group intelligence" that exists in electronic health records. Specifically, we measured laboratory test "repeat intervals", defined as the amount of time it takes for a physician to repeat a test that was previously ordered for the same patient. Our assumption is that while the result of a test is a direct measure of one marker of a patient's health, the physician's decision to order the test is based on multiple factors including past experience, available treatment options, and information about the patient that might not be coded in the electronic health record. By examining repeat intervals in aggregate over large numbers of patients, we show that it is possible to 1) determine what laboratory test results physicians consider "normal", 2) identify subpopulations of patients that deviate from the norm, and 3) identify situations where laboratory tests are over-ordered. We used laboratory tests as just one example of how physician group intelligence can be used to support evidence based medicine in a way that is automated and continually updated.
... 20 Many similar techniques have been used successfully since then and work best when the majority of the laboratory data come from a single homogeneous population that is therefore presumably unaffected by disease. [21][22][23][24][25][26] The challenge is to know when there are overlapping populations 27 and to have the understanding to know when to 'partition' that data into homogenous groups. 28,29 ...
... The normal value of a clinical measurement is usually defined by Gaussian distribution, which constitutes from the central 95% (or 2 standard deviations [SDs]) value of the healthy population [16,35]. We referred to data from several publications to estimate the normal reference range of human lung weight36373839. ...
Gravimetric validation of single-indicator extravascular lung water (EVLW) and normal EVLW values has not been well studied in humans thus far. The aims of this study were (1) to validate the accuracy of EVLW measurement by single transpulmonary thermodilution with postmortem lung weight measurement in humans and (2) to define the statistically normal EVLW values.
We evaluated the correlation between pre-mortem EVLW value by single transpulmonary thermodilution and post-mortem lung weight from 30 consecutive autopsies completed within 48 hours following the final thermodilution measurement. A linear regression equation for the correlation was calculated. In order to clarify the normal lung weight value by statistical analysis, we conducted a literature search and obtained the normal reference ranges for post-mortem lung weight. These values were substituted into the equation for the correlation between EVLW and lung weight to estimate the normal EVLW values.
EVLW determined using transpulmonary single thermodilution correlated closely with post-mortem lung weight (r = 0.904, P < 0.001). A linear regression equation was calculated: EVLW (mL) = 0.56 × lung weight (g) - 58.0. The normal EVLW values indexed by predicted body weight were approximately 7.4 ± 3.3 mL/kg (7.5 ± 3.3 mL/kg for males and 7.3 ± 3.3 mL/kg for females).
A definite correlation exists between EVLW measured by the single-indicator transpulmonary thermodilution technique and post-mortem lung weight in humans. The normal EVLW value is approximately 7.4 ± 3.3 mL/kg.
UMIN000002780.
... PAT and PAT/ET detect mice with elevated RVSP-In the present study, C57BL6 wild type mice had a RVSP of 27±2.5 mmHg. Elevated RVSP was defined as RVSP above the mean +2SD of the normal values (27+5), 19 yielding a threshold value for elevated RVSP of 32 mmHg. The hemodynamic and echocardiographic characteristics of the mice with normal or elevated RVSP are shown in Table 2. ...
Genetically modified mice offer the unique opportunity to gain insight into the pathophysiology of pulmonary arterial hypertension. In mice, right heart catheterization is the only available technique to measure right ventricular systolic pressure (RVSP). However, it is a terminal procedure and does not allow for serial measurements. Our objective was to validate a noninvasive technique to assess RVSP in mice.
Right ventricle catheterization and echocardiography (30-MHz transducer) were simultaneously performed in mice with pulmonary hypertension induced acutely by infusion of a thromboxane analogue, U-46619, or chronically by lung-specific overexpression of interleukin-6. Pulmonary acceleration time (PAT) and ejection time (ET) were measured in the parasternal short-axis view by pulsed-wave Doppler of pulmonary artery flow. Infusion of U-46619 acutely increased RVSP, shortened PAT, and decreased PAT/ET. The pulmonary flow pattern changed from symmetrical at baseline to asymmetrical at higher RVSPs. In wild-type and interleukin-6-overexpressing mice, the PAT correlated linearly with RVSP (r(2)=-0.67, P<0.0001), as did PAT/ET (r(2)=-0.76, P<0.0001). Sensitivity and specificity for detecting high RVSP (>32 mm Hg) were 100% (7/7) and 86% (6/7), respectively, for both indices (cutoff values: PAT, <21 ms; PAT/ET, <39%). Intraobserver and interobserver variability of PAT and PAT/ET were <6%.
Right ventricular systolic pressure can be estimated noninvasively in mice. Echocardiography is able to detect acute and chronic increases in RVSP with high sensitivity and specificity as well as to assess the effects of treatment on RVSP. This noninvasive technique may permit the characterization of the evolution of pulmonary arterial hypertension in genetically modified mice.
... Using the technique of Shine [31], another indirect procedure for the estimation of RLs, the Shine procedure led to higher values which were not plausible. Apparently, the Shine method is more sensitive to higher disease prevalences which can be expected in the older age groups. ...
Background:
The current dogma of establishing intra-laboratory reference limits (RLs) and their periodical reviewing cannot be fulfilled by most laboratories due to the expenses involved. Thus, most laboratories adopt external sources for their RLs often neglecting the problems of transferability. Therefore, several attempts were undertaken to derive RLs from the large data pools stored in modern laboratory information systems. These attempts were further developed to a more sophisticated indirect procedure. The new model can be considered a combined approach because it pre-excludes some subjects by direct criteria. In the current study, the new concept was applied to estimate RLs for serum and plasma creatinine from several German and Italian laboratories.
Methods:
A smoothed kernel density function was estimated for the distribution of the total mixed data of the sample group (combined data of non-diseased and diseased subjects). It was assumed that the "central" part of the distribution of all data represents the non-diseased ("healthy") population. The central part was defined by truncation points using an optimisation method, and was used to estimate a Gaussian distribution of the values of presumably non-diseased subjects after Box-Cox transformation of the empirical data. This distribution was now considered as the distribution of the non-diseased subgroup. The percentiles of this parametrical distribution were calculated to obtain RLs.
Results:
RLs determined by the indirect combined decomposition technique led to similar RLs as the classical direct method. Furthermore, the RLs obtained from 14 laboratories in 2 different European regions reflected the well-known differences of various analytical procedures. Stratification for gender and age was necessary. With rising age, an increase of the upper RL and of the reference range was observed. Hospitalization appeared also to affect the RLs. The new approach led to RLs in an artificially mixed population of diseased and non-diseased subjects (selected by clinical criteria) which were identical to RLs determined by a direct method applied to the non-diseased subgroup.
Conclusions:
The proposed strategy of combining exclusion criteria with a resolution technique led to plausible retrospective RLs from intra-laboratory data pools for creatinine. Differences between laboratories were mainly due to the well-known bias of the different analytical procedures.
... Using the technique of Shine [31], another indirect procedure for the estimation of RLs, the Shine procedure led to higher values which were not plausible. Apparently, the Shine method is more sensitive to higher disease prevalences which can be expected in the older age groups. ...
... In recent issues of the Annals this technique has been valuably used to develop reference ranges for children 4 and across the entire age range for serum alkaline phosphatase. 5 A further approach is being developed in the UK inline with our National Health Service. Currently, each laboratory has its own analytical methods and reference ranges with variations between neighbouring laboratories. ...
Background:
Due to different growth and metabolic processes, reference values of alkaline phosphatase (AP) for children aged 3 month to 18 years are dependent on age and sex. They are not constant and differ from those of adults due to the growth processes taking place. Accordingly, reference levels of AP continuous across these ages were generated for boys and girls based on of a large German health- and population-based study, LIFE Child. We considered AP at different growth and Tanner stages and additionally its association with other anthropometric parameters. The association between AP and BMI was of particulary great interest due to controversial literature on this topic. The role of AP in liver metabolism was investigated by examining ALAT, ASAT, and GGT.
Methods:
3976 healthy children (12,093 visits) were included from the LIFE Child study from 2011 to 2020. The subjects´ age ranged from 3 months to 18 years. Serum samples from 3704 subjects (10,272 cases, 1952 boys and 1753 girls) were analysed for AP after applying specific exclusion criteria. After calculating of reference percentiles, associations between AP and height-SDS, growth velocity, BMI-SDS, Tanner stage and the liver enzymes ALAT, ASAT and GGT were examined via linear regression models.
Results:
In the continuous reference levels, AP showed a first peak during the first year of life, followed by a plateau at a lower level until the start of puberty. In girls, AP increased beginning at the age 8, with a peak around 11 years, in boys beginning at the age 9, with a peak around age 13. Afterwards, AP values decreased continuously until age 18. In Tanner stages 1 and 2, AP levels did not differ between the two sexes. We found a strong positive association between AP-SDS and BMI-SDS. We also observed a significantly positive association between AP-SDS and height-SDS, which was stronger in boys than in girls. We found different intensities in the associations of AP with growth velocity depending on age group and sex. Furthermore, we found a significantly positive association between ALAT and AP in girls but not in boys, whereas ASAT-SDS and GGT-SDS were significantly positively associated with AP-SDS in both sexes.
Conclusion:
Sex and age, but also BMI may act as confounding factors for AP reference ranges. Our data confirm the remarkable association between AP and growth velocity (or height-SDS, respectively) during infancy and puberty. In addition, we were able to specify the associations between AP and ALAT, ASAT, and GGT and their differences in both sexes. These relations should be considered when evaluating liver and bone metabolism markers, especially in infancy.
Background and aims
Investigations in pregnancy should be interpreted using pregnancy-specific reference intervals (RIs). However, because of the progressive nature of pregnancy, even pregnancy-specific RIs may not be equally representative at different gestations. We proposed that gestational age-specific RIs may increase diagnostic accuracy over those with fixed limits.
Materials and methods
The trajectory of platelets was mapped in 32,778 pregnant women, using 116,798 results. Then we evaluated the accuracy with which a low measurement in early pregnancy (<3rd centile) predicted thrombocytopaenia at term, compared to the existing limit (<150 x10⁹/L).
Results
Platelets fell by 14.8% between 8-40 weeks. Platelets below the 3rd centile before 20 weeks predicted thrombocytopaenia at term (<100 x10⁹/L) with a significantly greater degree of accuracy than a fixed limit (AUC 0.86 vs. 0.76, p=0.004).
Conclusion
Pregnancy-specific RIs can be defined using routinely collected hospital data, and the abundance of such freely available data enables a detailed investigation of temporal changes throughout gestation. Individualised RIs offer improved accuracy profiles, over and above those already derived specifically from pregnant populations. Clinicians should consider how this may be used to improve diagnostic accuracy for biomarkers used in current clinical practice, and those yet to be defined.
The log transformation is often used to reduce skewness of a measurement variable. If, after transformation, the distribution is symmetric, then the Welch t-test might be used to compare groups. If, also, the distribution becomes close to normal, then a reference interval might be determined.
Background:
Interpretation of alkaline phosphatase activity in children is challenging due to extensive changes with growth and puberty leading to distinct sex- and age-specific dynamics. Continuous percentile charts from birth to adulthood allow accurate consideration of these dynamics and seem reasonable for an analyte as closely linked to growth as alkaline phosphatase. However, the ethical and practical challenges unique to pediatric reference intervals have restricted the creation of such percentile charts, resulting in limitations when clinical decisions are based on alkaline phosphatase activity.
Methods:
We applied an indirect method to generate percentile charts for alkaline phosphatase activity using clinical laboratory data collected during the clinical care of patients. A total of 361,405 samples from 124,440 patients from six German tertiary care centers and one German laboratory service provider measured between January 2004 and June 2015 were analyzed. Measurement of alkaline phosphatase activity was performed on Roche Cobas analyzers using the IFCC's photometric method.
Results:
We created percentile charts for alkaline phosphatase activity in girls and boys from birth to 18 years which can be used as reference intervals. Additionally, data tables of age- and sex-specific percentile values allow the incorporation of these results into laboratory information systems.
Conclusions:
The percentile charts provided enable the appropriate differential diagnosis of changes in alkaline phosphatase activity due to disease and changes due to physiological development. After local validation, integration of the provided percentile charts into result reporting facilitates precise assessment of alkaline phosphatase dynamics in pediatrics.
The national standard of laboratory quality control is based on a assumption about normal distribution of random error. The data of procedure of quality control get from analyzer Flexor XL during three months period were examined. The distributions of random error were non-Gauss ones in almost half of all examined tests by reason of both abnormal asymmetry and excess. To prevent possible clinical mistakes the actual level of quality is to be evaluated using dynamic values of bias, variation coefficient and o-metrics based on simple consecutive means. The most appropriate time of averaging consisted about 10 days and had no dependencies of type of test. The actual standard cab y easily improved taking into account the discovered advantages and simplicity of proposed approach.
Paediatric reference intervals are less well characterized than in adults. An initiative for harmonization of pathology across the United Kingdom has recommended an interval for sodium of 133-146 mmol/L at all ages.
To assess the validity of this, the laboratory database was interrogated for all renal profiles (sodium, potassium, urea and creatinine) for children presenting to primary care over a 13-year period. While the primary interest was in sodium results, sufficient current data were also available for potassium and creatinine and so these were included for study. The electrolyte results were filtered to include only normal renal function and the remaining data were analysed for age-related differences.
Sodium concentrations were observed to be lower for infants (1-5 years of age) with a mean of 138 mmol/L, increasing towards adult concentrations (mean 140 mmol/L) by teenage years. A similar pattern was seen for potassium results, and creatinine was seen to increase with age. At all ages, the distributions of sodium concentrations measured in this population were observably tighter than the interval of 133-146 mmol/L recommended by Pathology Harmony.
We suggest that this interval is too wide, and more work is needed to establish more appropriate paediatric ranges.
Age-specific paediatric reference intervals are used in interpretation of laboratory results. However, interpretation may be problematic when a child just crosses an age bracket and the difference between the original and the subsequent age-specific reference interval is large. Moreover, details about the physiological changes with age may be masked.For the 12 months ending 30 September 2013, results of 16 common clinical biochemistry tests of ambulatory paediatric patients aged 0-19, requested by primary care physicians, were retrospectively collected in a large pathology service, and used to construct smoothed centile charts using a penalised maximum likelihood method.From the developed centile charts, the concentrations of sodium, bicarbonate, creatinine, urate, total protein, and albumin all increased with increasing age of the children. In contrast, the concentrations of potassium, chloride, anion gap, calcium, phosphate and lactate dehydrogenase decreased with increasing age of the children. Changes in the concentrations of urea, alkaline phosphatase, glucose, and total cholesterol varied by age. Generally, the boys and girls shared similar trend patterns until 10-15 years of age, when variations in the age of onset of puberty and development caused the trends of some biochemical measures to differ.The paediatric biochemistry centile charts are intuitive tools to use. They complement age-specific reference intervals in the tracking, interpretation and discussion of laboratory results. They also enhance the understanding of underlying physiological changes in biochemistry in children.
To ascertain whether or not the change in blood collection tubes for plasma glucose from Fluoride/Oxalate to Citrate/Fluoride/EDTA has had an effect upon the glucose results.
Plasma glucose results from fasting patients from 2007 to 2012 were extracted from the laboratory information system. The data was evaluated in order to see the potential impact on patient results due to the change in glucose stabilizer implemented in September 2010.
The mean glucose result was increased by approximately 14 % (difference: +0.80mmol/L) after the implementation of the citrate-buffered tubes (mean value=6.45mmol/L; n=15 125) as compared to fluoride/oxalate tubes (mean value=5.65mmol/L; n=15 867).
An increase in glucose results is seen after changing to citrate-buffered tubes due to the improved stabilizing effect as compared to fluoride. Properly collected blood samples will lead to the patient being correctly diagnosed. However, decision limits and reference intervals for glucose may need to be revised using citrate-tubes.
The Stockholm Hierarchy is a professional consensus created to define the preferred approaches to defining analytical quality. The quality of a laboratory measurement can also be classified by the quality of the limits that the value is compared with, namely reference interval limits and clinical decision limits. At the highest level in the hierarchy would be placed clinical decision limits based on clinical outcome studies. The second level would include both formal reference interval studies (studies of intra and inter-individual variations) and clinical decision limits based on clinician survey. While these approaches are commonly used, they require a lot of resources to define accurately. Placing laboratory experts on the third level would suggest that although they can also define reference intervals by consensus, theirs aren't as well regarded as clinician defined limits which drive clinical behaviour. Ideally both analytical and clinical considerations should be made, with clinicians and laboratorians both having important information to consider. The fourth level of reference intervals would be for those defined by survey or by regulatory authorities because of the focus on what is commonly achieved rather than what is necessarily correct. Finally, laboratorians know that adopting reference limits from kit inserts or textbook publications is problematic because both methodological issues and reference populations are often not the same as their own. This approach would rank fifth and last. When considering which so called 'common' or 'harmonised reference intervals' to adopt, both these characteristics and the quality of individual studies need to be assessed. Finally, we should also be aware that reference intervals describe health and physiology while clinical decision limits focus on disease and pathology, and unless we understand and consider the two corresponding issues of test specificity and test sensitivity, we cannot assure the quality of the limits that we report.
Background:
Interventions to increase brown adipose tissue (BAT) volume and activation are being extensively investigated as therapies to decrease the body weight in obese subjects. Noninvasive methods to monitor these therapies in animal models and humans are rare. We investigated whether contrast ultrasound (CU) performed in mice could detect BAT and measure its activation by monitoring BAT blood flow. After validation, CU was used to study the role of uncoupling protein 1 and nitric oxide synthases in the acute regulation of BAT blood flow.
Methods and results:
Blood flow of interscapular BAT was assessed in mice (n=64) with CU by measuring the signal intensity of continuously infused contrast microbubbles. Blood flow of BAT estimated by CU was 0.5±0.1 (mean±SEM) dB/s at baseline and increased 15-fold during BAT stimulation by norepinephrine (1 µg·kg(-1)·min(-1)). Assessment of BAT blood flow using CU was correlated to that performed with fluorescent microspheres (R(2)=0.86, P<0.001). To evaluate whether intact BAT activation is required to increase BAT blood flow, CU was performed in uncoupling protein 1-deficient mice with impaired BAT activation. Norepinephrine infusion induced a smaller increase in BAT blood flow in uncoupling protein 1-deficient mice than in wild-type mice. Finally, we investigated whether nitric oxide synthases played a role in acute norepinephrine-induced changes of BAT blood flow. Genetic and pharmacologic inhibition of nitric oxide synthase 3 attenuated the norepinephrine-induced increase in BAT blood flow.
Conclusions:
These results indicate that CU can detect BAT in mice and estimate BAT blood flow in mice with functional differences in BAT.
The changes in androgen levels after the menopause and the effects of hormone replacement therapy (HRT) itself and the mode of HRT have not been established. The objectives of this study were to document the effect of age on androgen levels in a normal population of postmenopausal women and to investigate the effect of oral or transdermal HRT on androgen, oestradiol and sex hormone binding globulin (SHBG) values.
A cross-sectional study was conducted on 182 postmenopausal females aged 45-100 years, randomly recruited from the community. Serum levels of dehydroepiandrosterone sulphate (DHEAS), androstenedione, testosterone, free androgen index (FAI) and SHBG in a reference group of women, subdivided by age, menopausal status, HRT replacement and mode of HRT replacement.
Age was a significant factor affecting androstenedione (<55, > or =55 years) and DHEAS (<65, > or =65 years). HRT status was found to influence values for androstenedione, SHBG and FAI. Neither HRT nor age significantly influenced testosterone.
The observations suggest the need to provide age-specific reference intervals for androstenedione and DHEAS but not for testosterone and SHBG. The significant effect of HRT treatment would also suggest a need for treatment associated reference intervals for androstenedione, SHBG and FAI. Suggested reference intervals for each of these partitioned groups are presented.
: In a paper written in 1964, Box and Cox described a method for estimating transformations and showed how in suitable cases valuable increases in simplicity and efficiency were possible. Since that time, this technique has enjoyed wide practical use and considerable success. However, a recent theoretical paper by Bickel and Doksum (1981) seems to suggest that serious dangers are associated with the employment of this method, and speaks of 'instability' and 'cost' of estimation of the transformation. These difficulties seem to be associated with (1) examples which common sense would rule out, namely situations where the effect of transformation on the data is almost linear, so that it is a matter of indifference which transformation is used; (2) the idea that it makes sense to state conclusions in terms of a number measured on an arbitrary scale; (3) failure to take proper account of the Jacobian of the transformation.
Improvement in reference interval estimation using a new outlier detection technique, even with a physician-determined healthy sample, is examined. The effect of including physician-determined nonhealthy individuals in the sample is evaluated.
Traditional data transformation coupled with robust and exploratory outlier detection methodology were used in conjunction with various reference interval determination techniques. A simulation study was used to examine the effects of outliers on known reference intervals. Physician-defined healthy groups with and without nonhealthy individuals were compared on real data.
With 5% outliers in simulated samples, the described outlier detection techniques had narrower reference intervals. Application of the technique to real data provided reference intervals that were, on average, 10% narrower than those obtained when outlier detection was not used. Only 1.6% of the samples were identified as outliers and removed from reference interval determination in both the healthy and combined samples.
Even in healthy samples, outliers may exist. Combining traditional and robust statistical techniques provide a good method of identifying outliers in a reference interval setting. Laboratories in general do not have a well-defined healthy group from which to compute reference intervals. The effect of nonhealthy individuals in the computation increases reference interval width by approximately 10%. However, there is a large deviation among analytes.
The techniques of exploratory data analysis include a resistant rule for identifying possible outliers in univariate data. Using the lower and upper fourths, FL and FU (approximate quartiles), it labels as “outside” any observations below FL − 1.5(FU — FL) or above FU + 1.5(FU — FL). For example, in the ordered sample −5, −2, 0, 1, 8, FL = −2 and FU = 1, so any observation below −6.5 or above 5.5 is outside. Thus the rule labels 8 as outside. Some related rules also use cutoffs of the form FL — k(FU — FL) and FU + k(FU — FL). This approach avoids the need to specify the number of possible outliers in advance; as long as they are not too numerous, any outliers do not affect the location of the cutoffs.To describe the performance of these rules, we define the some-outside rate per sample as the probability that a sample will contain one or more outside observations. Its complement is the all-inside rate per sample. We also define the outside rate per observation as the average fraction of outside observations. For Gaussian data the population all-inside rate per sample (0) and the population outside rate per observation (.7%) substantially understate the corresponding small-sample values. Simulation studies using Gaussian samples with n between 5 and 300 yield detailed information on the resistant rules. The main resistant rule (k = 1.5) has an all-inside rate per sample between 67% and 86% for 5 ≤n ≤ 20, and corresponding estimates of its outside rate per observation range from 8.6% to 1.7%.Both characteristics vary with n in ways that lead to good empirical approximations. Because of the way in which the fourths are defined, the sample sizes separate into four classes, according to whether dividing n by 4 leaves a remainder of 0, 1, 2, or 3. Within these four classes the all-inside rate per sample shows a roughly linear decrease with n over the range 9 ≤ n ≤ 50, and the outside rate per observation decreases linearly in 1/n for n ≥ 9.A more theoretical approximation for the all-inside rate per sample works with the order statistics X(1) ≤ … ≤ X(n). In this notation the fourths are X(f) and X(n + 1 — f) with f = ½[(n + 3)/2], where [·] is the greatest-integer function. A sample has no observations outside whenever {X(f)−X(1)}/{X(n+1-f)−X(f)}≤k and {X(n)−X(n+1-f)}/{X(n+1-f)−X(f)}≤k. We first approximate the numerators and denominator in these ratios by constant multiples of chi-squared random variables with the same mean and variance. We then approximate the logarithm of each ratio by a Gaussian random variable, and we calculate the correlation between these variables from the fact that the ratios have the same denominator. Finally, a bivariate Gaussian probability calculation yields the approximate all-inside rate per sample. The error of the result relative to the simulation estimate is typically from 1% to 2% for 5 ≤ n ≤ 50.To provide an indication of how the two rates behave in alternative “null” situations, the simulation studies included samples from five heavier-tailed members of the family of h distributions. For a given sample size, the all-inside rate per sample decreases as the tails become heavier, and the outside rate per observation increases.
Age-specific reference intervals are commonly used in the routine monitoring of individuals, where interest lies in the detection of extreme values, possibly indicating abnormality. Here, a review is given of the wide range of statistical techniques which have been proposed for the construction of these intervals and issues such as the estimation of confidence bands and goodness of fit are discussed. Three methods, thought to be the most widely applied approaches, are considered in more detail. Comparisons are made on the basis of reference interval estimation for three real data sets.
Refence centile curves show the distribution of a measurement as it changes according to some covariate, often age. The LMS method summarizes the changing distribution by three curves representing the median, coefficient of variation and skewness, the latter expressed as a Box-Cox power. Using penalized likelihood the three curves can be fitted as cubic splines by non-linear regression, and the extent of smoothing required can be expressed in terms of smoothing parameters or equivalent degrees of freedom. The method is illustrated with data on triceps skinfold in Gambian girls and women, and body weight in U.S.A. girls.
Discharge diagnoses provide a possibility to select patients individually and then to establish reference values for both "pathological" and control groups. Currently, the available diagnostic information is still at its infancy and should be carefully evaluated before the reference values based on those groups are utilized. It is anticipated that electronic storage of diagnostic and therapeutic information will be applied more commonly in the future as the development of computers makes it easier. The advanced utilization of laboratory data challenges physicians both in the clinical and laboratory side to participate in this development in order to make the information systems serve their actual needs more closely.